Climatic implications of D/H ratios of meteoric water over North America (9500–22,000 B.P.) as inferred from ancient wood cellulose C-H hydrogen

δD and δ^(13)C values have been measured for the unexchangeable hydrogen and the total carbon of cellulose extracted from 40 North American ^(14)C-dated trees that range in age from 9500 to 22,000 years B.P. Meteoric waters which precipitated over ice-free regions of North America in the interval 14,000–22,000 B.P. had more positive δD values than corresponding modern waters by an average of 19‰. Lower ocean temperatures and smaller temperature gradients than exist at present between ocean and ice-free North America are indicated for the late Wisconsin glacial maximum. This is compatible with warmer winters and cooler summers for this glacial period. The δD value of the North American ice sheet during the Late Wisconsin maximum was approximately −100‰ as determined from the inferred δD values of the waters of proglacial lakes Agassiz and Whittlesey. From this figure the increase in δ^(18)O of the oceans during the glacial maximum can be calculated to have been +0.8‰. At the point where they began to move over the ice, air masses supplying moisture to the North American ice sheet contained a little more than 50% of their original moisture content, which is a much greater percentage than exists in air masses supplying the modern Greenland and Antarctic ice sheets. This relatively vapor-rich air coupled with lower summer temperatures, which reduced ablation, probably contributed to the maintenance and growth of the ice sheet. The transition from glacial to interglacial conditions on North America was rapid and occurred within a 2000–3000-year interval. However, the transition may not have been synchronous over North America. A 40-year δD record in a spruce branch from the Two Creeks (Wisconsin) forest (∼11,800 B.P.) shows large variations which suggest an unusual hydrologic environment in the area of the tree. Cellulose δ^(13)C values range between −20.8 and −25.9‰, but do not correlate with δD variations for the samples analyzed in this work. Thus, climatic significance of δ^(13)C variations cannot be resolved from these data